Close call on PSU Caps ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm building a PSU for a PA amp based on Mr Holden's asymetrical n-channel design ( somewhat modified for 10 output devices ).
The PSU rails are +/-70V , and I had trouble finding suitable reservoir caps within a sensible proce range , I'm simply NOT going to through 10,000uF at each rail and say 'hey it's only a PA amp..' .

the question I have is - If I use caps with a working voltage rating of 80Volts and add zener clamps to the rails to limit rail excursions would this be enough to protect the caps from re-decorating the inside of the case?

would the zeners be struggling to cope with the power spikes?, I was planning on using 5W 75V or 82V devices.

Each rail has 6 x 4700uF 80V caps, 2 x 100uF 100V caps and a couple of 0.1uF ceramics.

What would the likely rail excursions be driving a pair of 4R loads at about 400-500W ?
I don't have a simulator that can show me this sort of thing , and my 'scope is in for repair:( , so I can't even suck it and see:bawling:


any ideas?

Ray
 
TheFettler said:

The PSU rails are +/-70V , and I had trouble finding suitable reservoir caps within a sensible proce range , I'm simply NOT going to through 10,000uF at each rail and say 'hey it's only a PA amp..' .

the question I have is - If I use caps with a working voltage rating of 80Volts and add zener clamps to the rails to limit rail excursions would this be enough to protect the caps from re-decorating the inside of the case?

would the zeners be struggling to cope with the power spikes?, I was planning on using 5W 75V or 82V devices.

Each rail has 6 x 4700uF 80V caps, 2 x 100uF 100V caps and a couple of 0.1uF ceramics.

What would the likely rail excursions be driving a pair of 4R loads at about 400-500W ?

If the rails are 70v each, even using 75v caps would be fine. Why would you use zeners to limit rail excursions? This is not AC, but filtered DC. Except if your wall AC can go higher than that, that's different.

Then you should measure what's the maximum it can get to and use capacitors for that filtered DC. Zeners are not supposed to be used on a supply, except working much higher, and they might affect your audio.

I can't think of good sounding systems, except proper regulators, to prevent spikes problems. In fact spikes should be treated before they get to your supply, with a MOV or filter.

Don't use ceramics on your supply, only film caps as bypasses, polypropylene if possible.


Carlos
 
Exactly, if the rails are 70 V at NO load then the line variations might give an extra 10 % giving 77 V which is fine, IMO, as this is the exception case. The rail "excursions" in all other cases will be lower than 70 V, not higher. I would use them and feel like I did something right.
 
my concern for spikes is not those that originate from the mains input , rather the 'load dump' spikes you get from driving an inductive load , which would be coupled back into the PSU.

I figured that the caps *should* be ok - I mean I've seen Crest and Crown amp using 94volt rails and 100volt caps , but I couldn't remember whether they used any form of clamping to limit the voltage on the PSU caps.

I take the point on the ceramic caps BTW , I haven't actually fitted them yet, so I'll probably change them to film instead.

Ray
 
TheFettler said:
my concern for spikes is not those that originate from the mains input , rather the 'load dump' spikes you get from driving an inductive load , which would be coupled back into the PSU.

I figured that the caps *should* be ok - I mean I've seen Crest and Crown amp using 94volt rails and 100volt caps , but I couldn't remember whether they used any form of clamping to limit the voltage on the PSU caps.


If I'm not wrong they use switching supplies which are a different kind of beast. A spike from AC can go quite high at the output. They probably use some TRIAC protection or something, not simple zeners, and don't care so much for high quality audio.

And yes, their rails are quite high, but they design 1000W amps that will be handling all kinds of loads. They take quite a beat and keep going. Remarkable indeed. Though I don't know about their sound quality.

Former models seemed to sound fine, but I'm not too sure over the new ones.


Carlos
 
exploding electrolytics

your line voltage is specified for certain tolerances -- and the excursions can be wide. 15% to 20% is more like the allowance you should build in.

I may have mentioned this before -- I investigated a problem which Black and White photographers were having in our area -- I put a logging voltmeter on the computer and kept track for a couple 24 hour periods -- the excursion was from 109 VRMS to 128 VRMS, and these weren't just spikes -- but responses to load variation. (I call our electricity provider "Third World Electric"). I suppose it's somewhat less a problem in ss vs tubes.

As you are all aware, the response of photographic media is logarithmic so the problem people were having related directly to lamp output . I have also put my spectrum analyzer on the line -- this is truely instructive of why some people just can't get decent sound with all the crud which comes through.
 
The voltage ratings of capacitors are always derated from what they can actually withstand. Some manufactures carrry out this derating more than others, but most capacitors specified for power filtering should not be damaged by exposure to 10% and maybe even 20% or higher overvoltages for short periods of time. What really kills capacitors is ripple current and high temperature. You can use a 200 volt capacitor on a 100 volt line and if you exceed the ripple current rating, the cap will be destroyed.

So, the life of a capacitor is related to the voltage it sees, the ripple current, the operating temperature, and the particular characteristics of the capacitor as designed and manufactured. Many aluminum electolytics are specified for a particular operating life when operated at rated voltage, temperature and ripple current. Exceed any of these, and the life goes down exponentially. Operate below them, and life will be extended.

The short answer to your question, therefore, is that an 80V cap will work fine at 70 volts - provided the ripple current is not exceeded. If the line voltage goes up 20% for a short while and the cap sees 84V, no significant harm will result. If you want to maximize the life of your expensive capacitors, however, look at the data sheets for the capacitors you are thinking of using and see what impact voltage, temperature and ripple current have.

(There are some capacitor types in which the life is not extended by operating at a less than rated voltage. Consult the data sheets.)
 
Overvoltage causes greater leakage currents. This causes internal temperature rise. This in turn shortens their life greatly. A good quality cap from a reputable manufacturer will have a decent margin. Do yourself a favour and keep the electro's cool by placing them at the bottom of the psu so that the cool air gets to them first. Don't put them above hot heatsinks or transformers. Some electros are rated for 85 or even 105 deg C operation but they will only last about 2000 hours operation like this. I read somewhere in some cap manufacturer's literature that for every 10 deg C reduction in temp a cap will last 10 times longer. Keep it cool and you can make a good cap last decades while keeping it's specs.
 
I think you will be fine running ±70V with your 80V caps.

OTOH I think it would be very hard to get 500W/4R from this voltage. The Hafler DH500 is rated at 400W/4R (20hz~20Khz) and 600W/3R (1Khz) and runs on ±93V.

I would add a ±77V tier to power the voltage gain stages. This will add about 5V RMS to the output swing. 35V RMS/4R = 306W, 40V RMS/4R = 400W, in this case that extra 5V can make a 100W difference.

A 5-0-5V 20VA can be re-wired in series with each end of your 50-0-50V 1.5KVA transformer giving 55-50-0-50-55V AC and ±77V and ±70V after your rectifiers. A 1A bridge and a couple of 1000µF caps will be enough to run the front end.

The Hafler XL280, XL600, and the GAS Son of Ampzilla come to mind as models with high voltage tiers.
 
djk said:
I think you will be fine running ±70V with your 80V caps.

OTOH I think it would be very hard to get 500W/4R from this voltage. The Hafler DH500 is rated at 400W/4R (20hz~20Khz) and 600W/3R (1Khz) and runs on ±93V.

I would add a ±77V tier to power the voltage gain stages. This will add about 5V RMS to the output swing. 35V RMS/4R = 306W, 40V RMS/4R = 400W, in this case that extra 5V can make a 100W difference.


The Hafler uses MOSFET at the output, and MOSFETs seem to output less than bipolars, like a 10% less if I'm not wrong.


Carlos
 
Circlotron said:
Some electros are rated for 85 or even 105 deg C operation but they will only last about 2000 hours operation like this. I read somewhere in some cap manufacturer's literature that for every 10 deg C reduction in temp a cap will last 10 times longer.
Just a detail, but I believe that should have been "twice as long" (which I have seen in more thamn one datasheet).

/U.
 
If I have a 50V output voltage from a DC / DC converter, would I be okay with 50V electrolytics?

The particular caps I'm considering are the Nack series from Niccomp - their ultra low ESR series. They're rated for 63V surge I believe. Which I thought would cover any spikes in the output okay.

The next stage up is their 63V line, but they, obviously, use up more board space, and that is a very serious concern in this supply, as it needs to be portable.

Some of you may be interested in the Tanceram line offered by johansondielectrics.com. Whilst they have relatively low capacitance values, every voltage series offered by them is rated to run up to 160V and still operate! That's some serious derating! :D

They seem handy for things like switchmode supplies.

Also, I have carried out modelling using software from National Semiconductor that, regularly, recommends parts with a ripple current ability below that which will be present in the supply.

A lot of caps are rated for up to a hundred degrees plus, as has been mentioned. The ripple provides the self heating effect of the cap. If the ambient temperature is far below the caps rated value, the ripple current has less of an effect on lifespan. Remember that the ripple current is rated for the capacitor whilst it's running at the specified ambient temperature.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.